Background and purpose: Administrative registers, like hospital discharge registers and causes of death registers are used for the monitoring of disease incidences and in the follow-up studies. Obtaining reliable results requires that the diagnoses in these registers are correct and the coverage of the registers is high. The purpose of this study was to evaluate the validity of the Finnish hospital discharge registers and causes of death registers stroke diagnoses against the population-based FINSTROKE register.
Methods: All first stroke events from the hospital discharge registers and causes of death registers from the areas covered by the FINSTROKE register were obtained for years 1993-1998 and linked to the FINSTROKE register. The sensitivity and positive predictive values were calculated.
Results: A total of 3633 stroke events, 767 fatal and 2866 non-fatal strokes, were included in the registers. The sensitivity for all first stroke events was 85%, for fatal strokes 86% and for non-fatal strokes 85%. The positive predictive values for all first strokes was 86%, for fatal strokes 92% and for non-fatal strokes 85%. The sensitivity as well as the positive predictive values for subarachnoid haemorrhage and intracerebral haemorrhage was higher than for cerebral infarctions. There were no marked differences in the sensitivity or positive predictive values between men and women.
Conclusions: The sensitivity and the positive predictive values of the Finnish hospital discharge registers and causes of death registers are fairly good. Finnish administrative registers can be used for the monitoring of stroke incidence, but the number of cerebral infarctions should be interpreted with caution.